-
Notifications
You must be signed in to change notification settings - Fork 27.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Whisper] Pipeline: handle long form generation #35750
base: main
Are you sure you want to change the base?
Conversation
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's missing a test IMO! 🤗
elif self.type == "seq2seq_whisper" and not ignore_warning: | ||
logger.warning( | ||
"Using `chunk_length_s` with Whisper models is not recommended and will result in unreliable results, as it uses it's own chunking mechanism " | ||
"(cf. Whisper original paper, section 3.8. Long-form Transcription)." |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As I mentioned offline would be a pity to not use that batch algo in some cases! But up to debate!
What does this PR do?
Fixes #34210 #31942
In the tokenizer decoding logic for the pipeline, timestamp offsetting when the call to Whisper's generate have seeking (meaning generating for a new segment).
TODO
chunk_length_s=60
e.g. ? → actually Whisper just should not be used withchunk_length_s
set! Added a warning